How football predictions work
A football prediction is an estimate of how likely something is to happen in a match. This guide explains the idea in plain language, using FootIQ as the example.
Last updated:
From team strength to expected goals
Every model starts by estimating how strong each team is in attack and in defence, based on past matches. Combining the home team's attack with the away team's defence (and the other way round) gives the number of goals each side is expected to score in this match, for example 1.6 against 1.1.
From expected goals to probabilities
Goals in football follow a fairly predictable random pattern, so expected goals can be turned into the probability of every possible score: 0-0, 1-0, 1-1, 2-1 and so on. Adding up the right scores gives each market: all scores with two or more goals give "over 1.5 goals", all scores where both teams scored give "both teams to score".
Why several models
Different models make different simplifying assumptions. FootIQ uses three and weights them by how accurate each has been on matches it had not seen. When they strongly disagree, the pick is not published.
Calibration: making 70% mean 70%
A model can be systematically over- or under-confident. Calibration compares past predicted probabilities with what actually happened and corrects the scale, so that picks shown at 70% win about 70% of the time over many matches.
Why good predictions still lose
A 70% pick is expected to lose 3 times in 10. Over a handful of matches anything can happen; only over hundreds of predictions does the true accuracy become visible. That is why we show the sample size next to every rate on the track record.
Where to go next
The methodology describes FootIQ's exact approach, and prediction confidence explains how to read the numbers.